The 4 BACKTESTING TECHNIQUES behind WINNING Strategies: i've spent the last 2 years running backtests on everything from mean reversion setups to volatility arbitrage to prediction market signals some strategies survived and tbh most of them died and the difference was never the strategy itself, it was how i tested it a backtest is not proof your strategy works, it's a stress test to see how easily it breaks here are the 4 techniques i've actually run, what worked, what broke AND what i still use --------------- technique 1: standard in-sample / out-of-sample split verdict: broken by default, NEVER TRUST THIS the setup is easy, take 5 years of data and train on the first 4, test on the last 1 the problem is subtle - every time you tweak the strategy and re-run, you're peeking at the test data and after 30 iterations your "out-of-sample" is FULLY contaminated the first strategy i ever backtested was a simple pairs trade between two energy stocks that showed a Sharpe of 2.1 on the standard split, so i deployed $2,000 of my own money and lost 40% of it in 3 months going back later i realized i'd re-run that backtest 47 times during tuning, the test data was never really untouched use this only for a quick first look, NEVER as the final validation --------------- technique 2: walk-forward validation verdict: the real workhorse, this is what i actually use instead of splitting once, you slide a window through the data train on 2018-2020, test on 2021 train on 2019-2021, test on 2022 keep sliding each test window is data the model has never seen and you get 5 or 6 test periods instead of JUST ONE what this catches: > strategies that only worked in one regime (the pattern shows up immediately) > parameters that shift wildly when retuned (unstable strategy, red flag) > strategies that survive across every window (this is real edge) at our fund we killed a stat arb strategy that showed Sharpe 2.4 on a standard split, but walk-forward revealed it worked beautifully in 2019-2020 and completely died in 2021-2022, the regime had shifted underneath us and it saved us months of losses but this is slower and more painful than a standard split and it's also the reason institutional backtests match live P&L :) --------------- technique 3: purged k-fold cross-validation verdict: fixes a hidden bug in walk-forward financial data has memory, today's price is not independent of yesterday's when your training window ends on december 31 and your test window starts january 1, information leaks across that boundary and your Sharpe looks better than it should purged k-fold fixes this, Marcos Lopez de Prado covers it in Advances in Financial Machine Learning the idea is simple: > split data into folds like standard cross-validation > when a fold is used for testing, remove the adjacent observations that overlap in time > this eliminates the leakage a QUANT friend of mine who runs an ML-based factor model showed me his numbers before and after adding purging, Sharpe dropped from 1.9 to 1.4 on the same strategy with the same data and the extra 0.5 was pure leakage he didn't know he had use this when you're training ML models on financial data, the leakage in tree-based models is brutal without it --------------- technique 4: monte carlo trade shuffling verdict: the reality check that saves capital EVERY SINGLE TIME your backtest shows one sequence of trades, Monte Carlo randomizes the order and runs it thousands of times why this matters: > your backtest might have gotten lucky with sequencing, what if the drawdown happened in month 2 instead of month 10 > the max drawdown you observed is one path, Monte Carlo shows the full range > the 5th percentile drawdown is often 2 to 3 times worse than what you saw few months ago (during the hype of 15-min BTC markets) i built a systematic prediction market strategy that showed 12% max drawdown across 18 months of backtest, but before deploying i ran Monte Carlo with 10,000 shuffled sequences and the 5th percentile scenario showed a 34% drawdown ofc i didn't deploy at full size, i sized it at 25% of what i originally planned and three months in the strategy hit a 22% drawdown, but the smaller size meant i could hold through it and the strategy recovered to finish the year up 31% Monte Carlo is why i stayed in that trade instead of blowing up --------------- what i actually use in production NOW: > walk-forward validation as the primary test > purged k-fold when the strategy uses ML models > Monte Carlo shuffling on strategies that survive both, before any real capital > standard in-sample/out-of-sample only for the very first pass (rarely tho) if your backtest is designed to make you feel good, it's designed to LOSE you money for real.

The Clarity Act is America’s oppt to set the standard & bring innovation onshore. Capital and talent follow clear rules. Onchain finance and tokenization aren’t coming-they’re here. The question isn’t whether they’ll transform markets, but where. Don’t cede the future. Lead it. Thanks to @crypto for having me on to discuss the latest in the industry, including the upcoming vote.

The future of AI belongs to the humans behind it. There’s a common fear that as AI gets better, people get pushed out of the picture. We believe the opposite is happening. AI is creating entirely new categories of work, and an entirely new economy around the people whose knowledge, judgment, and experience are helping these systems improve. Today, tens of thousands of experts are actively contributing to AI training projects through micro1. Over time, we believe this will grow to tens of millions+. And if humans are going to play such a critical role in building the future of AI, the companies they work with should raise the standard for how they’re supported. Today, I’m proud to announce the micro1 Expert Support Program. We’re building a new set of protections, resources, and support for our expert community, starting with: -Expert Bill of Rights: a clear set of commitments outlining what experts can expect when working with micro1. -Rest Credits: paid time away from projects when experts need or want a break. -Expert Emergency Fund: financial support for experts facing emergencies. -Mental Health & Coaching: new resources to support expert wellbeing and growth. -Confidential Support Line: a confidential way to ask questions, seek support, or raise concerns related to pipelines. Within the next two weeks, every expert currently active with micro1 will receive an email with the full program details and launch dates. AI is going to keep getting more capable. The opportunity in front of us is to make sure the people helping build it benefit from that progress too.

THE MAG 7 WENT FLAT IN THE FIRST HALF OF 2026 AND THE S&P 500 GAINED 9.3% ANYWAY The other 493 stocks did that, and the trade built for exactly that outcome just crossed $100 billion. The Invesco S&P 500 Equal Weight ETF $RSP holds every company in the index at the same size, so no handful of giants can dominate it. It passed $100 billion in assets for the first time this month after more than $12 billion of inflows this year, per CNBC, and it is beating the standard S&P 500 by roughly 3% in 2026. The appetite comes down to concentration. Google $GOOGL, Amazon $AMZN, Apple $AAPL, Meta $META, Microsoft $MSFT, NVIDIA $NVDA and Tesla $TSLA are roughly a third of the regular index, so buying it means a third of your money rides seven stocks priced on the same AI theme. "You're not really picking a winning horse; you're betting on all the horses," said VettaFi research director Cinthia Murphy. The three biggest standard S&P 500 ETFs still hold close to $3 trillion combined.

The recently implemented Ethnic Unity and Progress Promotion Law of China provides a systematic legal framework for safeguarding the educational rights of young people in ethnic regions. #DeepChina# Article 15 of the Law stipulates that "the state comprehensively promotes the use of the standard spoken and written Chinese language nationwide" and that "the state respects and guarantees the learning and use of the spoken and written languages of ethnic minorities." The comprehensive promotion of the standard spoken and written #Chinese# language aims to equip every child with the ability to participate in broader national development; respecting and guaranteeing the learning and use of minority languages protects the rights and interests of all ethnic groups to preserve and pass on their linguistic and cultural heritage. The two objectives are mutually reinforcing. #EthnicUnity# The textbook system also enjoys clear #legal# protection. Article 16 of the Law explicitly stipulates that schools and other educational institutions at all levels and of all types shall use national unified textbooks in accordance with relevant regulations. The use of national unified Chinese language, history, and ethics and rule of law textbooks in classroom instruction shall help students develop a correct view of the state, history, ethnicity, and culture. China has never overlooked the fact that ethnic regions are home to numerous minority groups, including #Tibetan#, #Yi#, and #Uygur# communities, with rich and diverse languages and cultures. Consequently, ethnic regions have simultaneously introduced locally compiled textbooks and school-based reading materials, actively undertaking the function of preserving ethnic cultures. For example, Liangshan in Sichuan has organized the compilation and publication of over 1,200 types of Yi-language textbooks and readings, totaling some 120 million characters. #ChineseNation#

The insurance industry has priced the AI risk.  Carriers went to state regulators this year asking to exclude AI-related damages from ordinary general liability policies. 

Regulators approved more than 80% of those requests, and the standard forms behind roughly 82% of American property and casualty coverage now carry a generative AI exclusion that gets attached at renewal.  Underwriters are telling you they can't price a loss they can't inspect or audit . This is the same problem all large enterprises have with examiners, regulators or auditors when they ask “Show me what happened and demonstrate that your controls actually worked.” Imagine a bank builds an AI agent. The agent makes a decision that causes a $20M loss. The bank asks the insurer to cover it. The bank’s insurer asks: What exactly happened? Which model was running? What information did it receive? What tools did it invoke? What actions did it take? What rules constrained it? Was a human involved? Can you reproduce the sequence? If the answer is essentially “we don’t know - the model made the decision”, the bank has a nightmare and the insurer will balk. It can’t determine causality, negligence, controls, or even whether the same thing could happen tomorrow. But suppose the software has an immutable audit trail: Prompt → context → model → reasoning/action path → tool calls → data accessed → permissions → output → human approvals → final action Now the loss is inspectable. An insurer can underwrite it much more like conventional operational risk. Software that keeps a traceable record of what it did and why is how a regulated business answers these questions. This is the part we build:

The Panic of 1837 is one of the cleanest case studies you will ever find of government-manufactured financial catastrophe, and almost nobody has heard of it. Start with Andrew Jackson. He kills the Second Bank of the United States in 1832, which is the right call for the wrong reasons, because he hates central banking personally rather than on principle. Then his Treasury starts depositing federal funds into favored state "pet banks," which promptly use that hard-money base to pyramid credit on top of it. State-chartered banks across the South and West issue paper notes backed by essentially nothing, and land speculators borrow those notes to buy federal land at $1.25 an acre. Cotton prices are climbing. Everyone feels rich. The boom is artificial from the first dollar. Jackson then panics at his own creation. His Specie Circular of July 1836 mandates that buyers pay for federal land in gold or silver only. Credit evaporates overnight. Land prices collapse. Cotton follows, hitting around 9 cents per pound by 1837, down from 17 cents in 1835. Banks that had lent recklessly against inflated land values now hold collateral worth half what they financed. Over 600 banks suspend specie payments between May and October 1837. Nine states default on their bonds. Martin Van Buren inherits the wreckage and gets blamed for the explosion Jackson lit. The lesson sound money advocates have spelled this out since the 19th century: credit expansion without real savings does not create wealth, it relocates it forward in time and then destroys it. Every dollar of paper the pet banks printed above their specie reserves was a promise they could not keep. The speculative cotton and land mania of 1835-36 was the entirely rational response of market participants to artificially cheap credit. You would have done the same thing. Everyone did. What makes this maddening is the standard historical narrative: it blames "speculation" and sometimes blames Jackson's Specie Circular in isolation, as if tightening credit were the disease rather than the cure arriving too late. The disease was the credit expansion. The correction was painful precisely because the distortion was enormous. Governments do not create panics by suddenly demanding honest money. They create panics years earlier, when they let banks print paper out of thin air.